316 JOURNAL OF THE SOCIETY OF COSMETIC CHEMISTS the product's ability to pass such a test does not measure the overall hygienic production of the product or indeed the actual level of bacterial contamina- tion present. It was also felt to be necessary to incorporate into the stability programme regular microbiological analysis, so that knowledge could be gained of the microbiological life-span of products and their preservative systems and sufficient data acquired for meaningful co-operation with government bodies, dealing in practicalities rather than airy-fairy theories. All the QC managers were involved from the outset of the programme so that due economy in materials and time could be obtained. At the outset it was felt that it should be possible to introduce MQC to non-microbio- logically qualified staff if methods and the results obtained by those methods did not contain elements of judgment and opinion. By this it was meant that the staff and premises available at that time were sufficient to initiate and maintain a programme of microbiological testing without: (a) employing microbiologists in each plant (b) installing a separate microbiological laboratory in each plant. To this end a manual was prepared by qualified microbiologists in R & D which outlined methods for use in QC laboratories to control the final product and establish accept/reject limits. Over the course of the last 4 years the programme has grown to adulthood and each plant now has at least one person involved in final product analysis, 'in process' control, and monitor- ing general factory hygiene standards. These people in the majority of cases are QC chemists who took over the task because of personal interest, the minority are fundamentally more microbiologically orientated who were taken on in the course of natural turnover of staff in the QC laboratory. It is generally felt that the introduction was a challenge well met rather than a traumatic experience. As the author has said, elsewhere (1), 'Each manufacturer must assess and impose suitable standards for his product lines. Regrettably, standards will vary in an absolute sense throughout industry since the present structure of the cosmetic and toiletry industry interweaves with the manufacture of sterile and non-sterile pharmaceuticals and foodstuffs.' Throughout the plants regardless of product category the 'in house' standard imposed was, from the beginning, 100 organisms per g or per ml of product due to the fact that non-sterile pharmaceuticals and toiletties were and are manu- factured alongside each other. In the beginning all washing and disinfecting procedures were reviewed

MICROBIOLOGICAL QUALITY CONTROL 317 and in sensitive product areas were checked by R & D personnel. Only then was final testing of finished packed products introduced, product by product, with priority given to products known to be at risk in the factory (aqueous 'cold mix' products) and if contaminated those of direct risk to the con- sumer, for example, baby toiletries and oral non-sterile pharmaceuticals. The final check was by means of a total viable count per ml or gram of product, with the count carried out in such a way as to ensure the viability of as wide a range of organisms as possible and in the arithmetical treatment of numbers an inbuilt bias for counting high. In the early days of the programme no attempt was made by QC personnel to identify organisms, their role was purely that of quantification. If high counts were met, R & D personnel would visit the plant and after investigation pinpoint the source of contamination. In this way everyone, QC, Production and R & D learned from errors and mistakes. A plant with- out problems teaches nothing but merely adds to a sense of false security. The author has previously outlined the basic techniques for total viable count (1), but several points bear mention yet again. The aim of a viable count is to provide a series of conditions under which each viable microbial unit in a material will give rise to a discrete visible colony which can then be counted. It is easier in microbiology to determine sterility (under defined conditions) than it is to provide accurate assessment of counts between sterility and 100 organisms per g or ml. This is due to the fact that very few products could be plated directly without dilution and such dilution rapidly affected the reliability of the count. For example, with accept/reject limits at 100 a dilution of 1 in 10 will require a decision on counts of less than 10. This is an order of magnitude lower than the density usually accepted as optimal. It was fortunate that right at the beginning two temperatures of incuba- tion (24øC and 37øC) were used rather than one as in the US Pharmaco- poeia (2) (32øC). Bacteria such as Achromobacter which do not grow above 30øC have been a major problem over the years. Likewise, it was ensured that organisms with different growth requirements were catered for by using two non-specific growth media, nutrient agar for bacteria and Sabouraud dex- trose agar for fungi. With various products at one time or another, spread plates, pour plates, membrane filtration and most probable number methods have been used for viable count. Where a product produced in many centres has differing standards imposed by various governments, as a matter of good quality assurance the highest standard required has been imposed as the minimum company